Skill-based Task Execution on Skills-Bench 27-task headline sample
0.705Mean Task RewardAIP-compiled skills
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| AIP-compiled skillsSolver=Claude Sonnet, Trials per task=52026.06 | 0.705 | 67.4 | 510 | 25.6 | 12 | 13 | 2 | 0.011 | |
| Human-curated skillsSolver=Claude Sonnet, Trials per task=52026.06 | 0.599 | 53.3 | 585 | 27.7 | — | — | — | — |